Racing to commercialize AVs

Self-driving fleets? Idled. Plans to launch robotaxi services? Delayed. The future of ride-hailing and public transit? Suddenly in doubt. At least in the United States.

In China, a different story has materialized. Through the throes of the coronavirus crisis, Chinese companies developing autonomous-driving technology have accelerated the pace and scope of their deployments. Investors have rewarded those efforts, pumping hundreds of millions into their coffers at a time when such funding has been scarce elsewhere.

While most Chinese executives concede the likes of Waymo and Cruise still sit atop the list of global front-runners in the fledgling self-driving industry, they believe they're narrowing the gap.

Case in point: Robotaxi company WeRide received a permit Friday, July 10, from Chinese authorities to launch the country's first fully driverless tests on public roads in Guangzhou. Though it's unclear when the tests will commence, they would be on a par with Waymo, which has operated driverless Chrysler Pacifica minivans in a geofenced area near Phoenix.

Amid the pandemic, the broad Chinese advance has come as attention shifts from developing AV technology itself to focusing on practical operational considerations for fledgling robotaxi businesses.

"The Chinese industry is getting more mature," said Jianxiong Xiao, founder of AutoX. "In China, the history of autonomous driving is very short, four years. The first one, two, three, they needed to focus on the technology. We had to build that first. Now we have to think about how we commercialize."

AutoX has done more than think. In April, the company launched a fleet of 100 self-driving vehicles in Shanghai and opened an operations center that is a prominent self-driving vehicle data hub in China. At the same time, the company kicked off a partnership with mobility platform Amap, so that riders can use an app to arrange self-driving service — which still uses human safety drivers — or a ride in a human-driven vehicle.

In a development that showcases the increasing sophistication of the company's technology, AutoX can start or stop rides at any location requested within the company's geofenced operating area. Competitors must pick up and drop off passengers at specific stations.

Those are substantial achievements for a company founded four years ago in San Jose, Calif. But in many ways, they are mirrored by other Chinese companies that have commenced a flurry of activity in recent months. Among other key developments that have occurred as or shortly after the nation began to grapple with the coronavirus:

• In February, Pony.ai received $400 million in investment from Toyota Motor Corp., which brought the company's overall funding to $762 million. In April, Pony launched a delivery-oriented pilot project in Irvine, Calif.
• In May, Baidu completed construction on a 118,000-square-feet facility in Beijing, which houses 200 of the company's automated vehicles, an R&D center, maintenance operations, a security-command center and a closed-loop testing road. One month earlier, Baidu opened to the public its robotaxi service in Changsha, in Hunan Province, where it operates approximately 45 vehicles with human safety drivers.
• In May, ride-hailing network Didi Chuxing received $500 million from Japan's SoftBank Vision Fund, and later announced plans to deploy 1 million driverless vehicles on Chinese roads by 2030.

"In 2025, I think we're going to see mass-produced vehicles coming off production lines that don't have a steering wheel or brakes or mirror," said Xing Meng, COO of Didi's autonomous-driving unit, which was spun into an independent entity in August. "They're going to come out with an HD map installed off the line, and off it goes into a geofenced area."

Meng spoke during a panel discussion on China's full-speed-ahead autonomous mindset, hosted by the South China Morning Post.

He downplayed the idea of global competition. But his outlook aligns with Made In China 2025, a government-led policy written five years ago that set plans and policies that would push the nation to global leadership in high-technology industries, such as autonomous driving. The Chinese government anticipates that highly automated vehicles will make up 15 percent of sales in five years, and officials have been investing in smart infrastructure to help hasten their arrival.

"Commercialization is where the rubber meets the road," said Michael Dunne, CEO of ZoZo Go, a consulting firm that supports companies seeking to do business in Asia. "So we see Baidu, Pony, WeRide, AutoX moving very quickly with local governments to accelerate the commercialization of their technologies. And that's going to prove out who's the winner or loser."

The government set the blueprint. But China's autonomous-vehicle ambitions took root in California.

At least a dozen Chinese autonomous-vehicle companies hold permits to test on the state's public roads, according to California Department of Motor Vehicles records. The group includes established automakers such as SAIC Corp. and Changan; electric vehicle makers such as Nio; startups such as AutoX and WeRide, and under-the-radar companies such as Qcraft.ai.

Companies have relied on Silicon Valley's talent to jump-start their efforts and relied on California's roads as straightforward locations for teaching their vehicles self-driving basics. Now that their goals are shifting toward commercial operations, California's role may dwindle as companies seek areas with more chaotic traffic environments.

"Suppose you are playing basketball and practicing in the NBA," said Tony Han, CEO of WeRide. "Maybe that's more useful than practicing three years in college."

WeRide finds Chinese roadways more confounding — and advantageous. Lane changes are five times more frequent in Guangzhou than in Silicon Valley, and the company's cars are 60 times more likely to encounter cyclists riding against traffic, according to a company analysis.

Sending cars into that complexity may be financially lucrative. Consulting firm Bain and Co. estimates the Chinese ride-hailing market to be worth $23 billion, the largest in the world. Approximately 90 percent of that is accounted for by Didi, the firm said. But there's room for others.

Didi can meet 75 percent of today's demand. Meng said a quarter of all orders for rides are never picked up either because drivers are not nearby or there's simply not enough of them. Adding robotaxis to the fleet will alleviate some of that imbalance, he said.

Although it dropped precipitously at the height of the health crisis, ride-hailing demand is poised to grow as a ripple effect of the coronavirus, Xiao says.

At first glance, it may be difficult for Americans accustomed to car ownership to understand why COVID-19 would not be a hindrance. But he says ride-hailing could be preferred by residents of dense cities across Asia who typically take public transportation.

"More and more people prefer to have ride-hailing," Xiao said. "They don't want to share a ride with other strangers, but ride-hailing is becoming much more popular recently. And in a robotaxi, you can remove the driver and the car is completely private. That's even better."

While the perception that ride-hail vehicles are safer helps fuel demand, a key technology may help them stay on the road. Many Chinese companies have incorporated V2X technology into sensor suites.

Shorthand for "vehicle to everything," V2X enables cars to receive transmissions via a cellular connection from elsewhere in the road environment. Right now, traffic lights in certain cities can communicate signal and time-to-change information to self-driving cars, which can use that information in their decision-making processes.

It's a marked difference from the U.S., where most companies eschew the idea of relying on the government to install equipment that may be important to their near-term plans. V2X may still be a powerful safety tool for both human-driven and automated vehicles, but deployments in the U.S. have been mired in arguments over competing methods for transmitting messages.

Traffic-light information is already incorporated into many AVs in China today, and could play a larger safety role for autonomous vehicles later, helping to provide information that complements the sensors on board vehicles and reducing the number of confounding scenarios. V2X offers the prospect of enhancing the safety of all vehicles on the road, whether human driven or automated, by providing information beyond line of sight.

"It's a streetlight versus a headlight," Han said. "We have to make sure our cars are capable of autonomous driving without it. It doesn't replace anything we've developed over the past 10 years. But this will make autonomous driving more reliable."

Despite the momentum across China, a pragmatism remains in most of the executives. Han said he believes the Chinese companies are collectively a year or two behind their U.S. counterparts.

And there are undefined hurdles. Meng says the prospect of scaling from a fleet of several hundred test cars to several hundred thousand cars in operation is one that no company has truly faced yet, and much work remains in optimizing charging, refueling, maintenance, insurance and financing.

"One hundred cars, you can keep track on an Excel spreadsheet," he said. "But 500,000 cars, that's a huge asset-management problem."

Still, given the technology hurdles already overcome and the government support behind the broad autonomous endeavor, those are surmountable challenges, ones that the Chinese have taken the initiative to solve in recent months.

"It's happening," Dunne said. "The Chinese are always seemingly quicker to commercialize technology than we are. We're still the fount — the best technology — in California. But in terms of commercialization, it's very tough to beat the Chinese."